Breaking the next Cryo-EM resolution barrier – Atomic resolution determination of proteins!

Background

Cryo-electron microscopy (Cryo-EM) has become very popular and successful in solving three- dimensional (3D) structures of macromolecular complexes. The technological development of electron microscopes, detectors, automated procedures as well as friendly image processing software and increasing computational power, have made cryo-EM a successful and largely expanding technology over the last decade. At resolutions better than 4 Å, atomic model building starts becoming possible but the direct visualization of true atomic positions in protein structure determination requires significantly higher resolution, which so far could not be attained by cryo-EM. In their work Yip et al (1) present a newly developed electron microscope which provides an unprecedented resolution of 1.25 Å, and for the first time allows visualization of individual atoms in a protein (apoferritin). Moreover, their setup allows also significant improvements in quality of the cryo-EM density map.

Key findings and developments

The authors explored two key questions in their work, namely whether technical improvements in hardware would allow overcoming current optical limitations in Cryo-EM, and how far such technology can be pushed.

Hardware development

The authors equipped a Thermofisher Titan Krios electron microscope (with a Falcon 3 direct electron detector), with additional electron-optical elements to increase the performance of the microscope. These elements included amonochromator and a second- generation spherical aberration corrector. The monochromator aided in reducing the energy spread of the electron beam, while the aberration corrector allowing obtaining images free of axial and off-axial coma, as well as images free of other aberrations. This setup provided significantly increased temporal coherence and less dampening of high-resolution structural details in the images. The setup also minimized linear distortions, and remains stable over longer operation times. The images obtained from this new setup, therefore, do not require post-acquisition image processing to correct these aberrations. The authors point out, however, that in data collected over several months, a small change in overall magnification occurred which needed correction. Finally, the authors discuss that in order to achieve the highest possible resolution, an extremely high quality of biological specimens is required and therefore great care must be taken in sample preparation.

Achieved resolution and image quality

Imaged at low electron dose can result in images suffering from a significant amount of noise. To overcome this, an averaging procedure using several hundred thousand or even millions of particle images is used to calculate one 3D structure by image processing. How many such particle images are needed to obtain a specific resolution can be described by an experimental «B factor» (2). The setup hereby presented has an experimental B factor of 36 Å2. The authors predicted that using image processing alone, achieving a 1 Å resolution of apoferritin would require an unrealistic amount of time, computer power, and storage capacities. The authors went on to determine what are the particle numbers needed for the highest attainable resolution, and what features would become visible in cryo-EM maps at the highest obtained resolution with the instrument hereby presented. While an initial calculation indicated that over 5 million apoferritin particle images would be required to attain a resolution of 1.3 Å, optimization of grid preparation and imaging conditions allowed crossing the 1.5 Å resolution barrier with only 17.800 particle images. Eventually, a total number of 1.000.000 particle images allowed achieving 1.25 Å resolution. The obtained map achieved well-defined additional densities that agree with the positions of hydrogens on almost all atoms when using low thresholds for map visualization. The resolution of the map is sufficient to observe a sulfur chemical modification in apoferritin, which had not been visualized before.

In cryo-EM, the resolution of the map is estimated by the correlation of two independently calculated structures in various resolution shells in Fourier space (Fourier shell correlation (3)). The FSC provides a single number for the obtained resolution, but does not provide a direct measure of the quality of a 3D structure. Therefore, an independent means of map quality comparison is desirable. The authors compared a 1.55 Å resolution structure from a subset of the data acquired intheir instrument, with the highest resolution apoferritin structure at 1.54 Å (acquired with a different setup). They found that  seven times less data was necessary with their setup, and that although the nominal resolution was similar, a significant improvement in map quality was achieved with the setup hereby presented.

The authors then discuss improvements beyond their setup, which would be desirable in new EM hardware. Improvements discussed include a) improvement in image recording speed by faster cameras and optimized data acquisition schemes; b) next generation detectors which could potentially improve the “B factor” to <30 Å2 , needed to break the 1 Å resolution barrier with a manageable amount of data. Altogether, the authors conclude that improving electron microscope hardware is essential for Cryo-EM in terms of resolution and throughput.

 

What I like about this preprint

I chose this preprint because the achievement is completely novel. I think technology improvements have led the way and preceded many important and paradigm-shifting biological discoveries. I liked the questions the authors explored and the ones raised for the future. I enjoyed reading the preprint, and I think this achievement will have great impact in various disciplines.

Open questions

1. I enjoyed a lot reading your preprint. I was wondering if you could expand further (for experts and non-experts) on the applications for which your improved setup could be applied? It’s opening a whole new set of questions..
2. You mention in your preprint that despite the improvements your setup achieves, further improvement can be reached to achieve high throughput and higher resolution. Initially one of the questions you raised was whether improvements can come solely from the hardware side. Software developments seem to be taking place also at a high pace. Do you envisage a combination of hardware and software will allow the further achievements you discuss? If so, what are main factors to improve when it comes to software that you can jointly address with the hardware improvements you are exploring?
3. Regarding the further hardware improvements you discuss, can these be implemented in the setup you present here?

References

1. Yip KM, et al Breaking the next Cryo-EM resolution barrier- atomic resolution determination of proteins, bioRxiv, 2020.
2. Rosenthal, P. B. & Henderson, R. Optimal determination of particle orientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy. Journal of Molecular Biology 333, 721-745, 2003.
3. Harauz, G. & van Heel, M. Exact filters for general geometry three dimensional reconstruction. Optik (Stuttgart) 73, 146-156, 1986.

 

doi: https://doi.org/10.1242/prelights.22687

Read preprint (No Ratings Yet)